Download Exogenous Oil Supply Shocks

Exogenous Oil Supply Shocks:
How Big Are They and
How Much Do They Matter for the U.S. Economy?
Lutz Kilian
University of Michigan and CEPR
Motivation
● Wars and other exogenous political events in OPEC countries cause oil production shortfalls.
Examples: Iranian revolution (1978/79), Iran-Iraq War (1980-1988), Persian Gulf War (1990/91),
Iraq War (2003), Civil unrest in Venezuela (2002/03), and perhaps the Yom Kippur
War/Arab oil embargo (1973/74)
● Three key questions:
1. How large are the exogenous fluctuations in the production of oil?
2. What are the dynamic effects of exogenous oil production shortfalls on U.S. real GDP growth
and CPI inflation?
3. To what extent do exogenous oil supply shocks explain changes in the price of oil?
Part 1: Measuring Exogenous Oil Supply Shocks:
● Any attempt to identify the timing and magnitude of these exogenous production shortfalls
requires explicit assumptions about the counterfactual path of oil production in the absence of the
exogenous event in question.
● The strategy is to generate the counterfactual production level for the country in question by
extrapolating its pre-war production level based on the average growth rate of production in other
countries that are subject to the same global macroeconomic conditions and economic incentives, but
not affected by the war.
● Which countries belong into the benchmark group must be decided on a case-by-case basis
drawing on historical accounts and industry sources.
Example 1: Counterfactual for the
October 1973 War and the 1973/74 Arab Oil Embargo
● It is well known that oil production from Arab OPEC countries fell between September and
November of 1973, whereas oil production in the rest of the world did not. This observation suggests
that we take non-OPEC oil producers as our benchmark.
● Simply comparing the production decisions in non-OPEC countries and Arab-OPEC countries in
late 1973 would be misleading, however, because of differences in economic incentives across these
countries.
● 1971 Tehran/Tripoli agreements between the oil companies and Middle Eastern OPEC oil
producers:
- Duration of five years
- Moderate improvement in the financial terms that host governments received from oil companies
for each barrel of oil extracted by the oil companies …
- … in exchange for assurances that these governments would allow the oil companies to extract as
much oil as they saw fit on those terms.
The October 1973 War and 1973/74 Embargo: Production
2000
Crude Oil Production Relative to Non-OPEC Countries
Old Price Regime
New Price Regime
1500
1000 Barrels/Day
1000
500
0
-500
September 1973
-1000
-1500
-2000
Iraq
Iran
Saudi Arabia
Kuwait
Other Arab OPEC
January 1974
1974
March 1974
The October 1973 War and 1973/74 Embargo: Production Shortfall
3000
2000
Production Shortfall Associated with October War and Arab Oil Embargo
Old Price Regime
New Price Regime
1000 Barrels/Day
1000
0
-1000
September 1973
-2000
-3000
January 1974
-4000
1974
March 1974
Example 2: Iraq
Event:
Starting date:
Production Benchmark:
Iranian revolution
1978.10
Total OPEC – Iraq, Iran, Saudi Arabia
Iran-Iraq war
1980.10
Total OPEC – Iraq, Iran, Saudi Arabia
Persian Gulf War
1990.8
Total OPEC – Iraq, Iran, Saudi Arabia, Kuwait
Venezuelan Crisis
2002.12
Total OPEC – Iraq, Iran, Saudi Arabia, Kuwait, Venezuela
Iraq War
2003.03
Total OPEC – Iraq, Iran, Saudi Arabia, Kuwait, Venezuela
Iraq: Actual and Counterfactual Production
Crude Oil Production: Iraq
3500
3000
1000 Barrels/Day
Counterfactual
2500
2000
1500
1000
500
0
Actual
Iranian
Revolution
1978/79
1975
Iran-Iraq
War
1980-1988
1980
1985
Gulf
War
1990/91
1990
Iraq
War
2003
1995
2000
From Production Shortfalls to Supply Shocks
Procedure for Constructing a Historical Series of Exogenous OPEC Oil Supply Disturbances:
1. Sum all OPEC oil production shortfalls discussed so far (including the 1973/74 event which may
or may not have been exogenous).
2. Express this time series as a share of world crude oil output.
3. A natural measure of the exogenous OPEC oil supply shock is the change in the normalized
production shortfall over time.
Baseline Exogenous Oil Supply Shock Series for all of OPEC
First Difference of Exogenous Oil Production Shortfall: OPEC
5
Percent Share of World Production
Explicit Counterfactual
0
-5
Hamilton
-10
1975
1980
1985
1990
1995
2000
Part 2: The Dynamic Effects of Exogenous Oil Supply Shocks
● Let xt denote the date t observation of the exogenous oil supply shock series, Δyt the corresponding
percent growth rate in real GDP and π t the percent change in the consumer price index. The
ultimate object of interest are the impulse responses ∂Δyt +i ∂xt and ∂π t +i ∂xt , i = 1, 2,3...
● For each country, the first-order effect of a given increase in xt on Δyt +i and π t +i , respectively,
may be computed based on the fitted value of the linear ordinary least squares (OLS) regressions
4
8
4
8
i =1
j =1
i =1
j =1
Δyt = α + ∑ β i Δyt −i + ∑ γ j xt − j + ut and π t = δ + ∑ λiπ t −i + ∑η j xt − j + vt ,
where the error terms ut and vt are serially uncorrelated, given the inclusion of four lags of the
dependent variable and eight lags of the exogenous oil supply shock.
● Level responses for real GDP and the level of consumer prices may be obtained by cumulating the
estimated impulse responses. Confidence intervals for these impulse responses may be constructed
by drawing from the asymptotic normal distribution of the slope parameters and simulating the
standard errors of the impulse response estimators.
Dynamic Effects of a 10% World Oil Supply Disruption on United States
Direct Estimate Based on Baseline Counterfactual
Real GDP Level
1
0
-1
-2
0
-5
-10
-3
-4
1
Quarterly CPI Inflation
5
2
3
4
5
6
7
8
9
10
11
-15
1
12
3
15
2
10
1
5
CPI Level
Quarterly Real GDP Growth
2
0
-5
-2
-10
2
3
4
5
6
7
8
9
10
11
-15
1
12
Quarters
3
4
5
2
3
4
5
6
7
8
9
10
11
12
6
7
8
9
10
11
12
0
-1
-3
1
2
Quarters
5
Beyond dynamic multipliers …
There is a tendency to think of exogenous oil supply shocks as adverse
oil supply shocks. This need not be the case:
● Historically, exogenous production shortfalls have tended to be
temporary. In that case, by construction, negative shocks to oil
production are followed by positive shocks.
● In assessing the overall impact of exogenous oil supply shocks, it is
therefore more informative to conduct a counterfactual historical
simulation.
6
U.S. Real GDP Growth and CPI Inflation Relative to Long-Run
Average and Estimated Effect of Exogenous Oil Supply Shocks
Real GDP Growth
4
4
Real GDP Growth
4
Real GDP Growth
4
3
2
2
2
2
2
1
0
-1
-2
-4
1974
8
1974.5
1975
1
0
-1
CPI Inflation
0
-1
-2
-2
-3
-3
-4
1979
1975.5
1
8
1979.5
1980
-4
1981
1980.5
CPI Inflation
8
2002.IV-2004.III
3
1990.III-1993.III
3
1980.IV-1983.I
3
-3
1
0
-1
-2
-3
1982
-4
1983
CPI Inflation
0
-1
-2
-3
1991
1992
-4
2003
1993
CPI Inflation
8
8
7
6
6
5
5
5
5
5
3
2
1
0
4
3
2
1
0
4
3
2
1
0
2002.IV-2004.III
7
6
1990.III-1993.III
7
6
1980.IV-1983.I
7
6
4
4
3
2
1
0
0
-1
-1
-1
-2
-2
-2
-3
1979
1979.5
1980
1980.5
-3
1981
1982
1983
-3
Excess Relative to 1973.1-2004.III Average
Effect of Exogenous Oil Supply Shock
1991
1992
1993
CPI Inflation
1
-2
1975.5
2004.5
2
-1
1975
2004
3
-2
1974.5
2003.5
4
-1
-3
1974
Real GDP Growth
1
7
1978.IV-1980.III
1973.IV-1975.II
Real GDP Growth
3
1978.IV-1980.III
1973.IV-1975.II
4
-3
2003
2003.5
2004
2004.5
Part 3: The (Tenuous) Link from
Exogenous Oil Supply Shocks to Oil Price Shocks
1. There is widespread recognition today that oil prices since 1973 must be considered endogenous
with respect to global macroeconomic conditions.
2. Recently the case has been made that, nevertheless, nonlinear transformation of the price of oil
designed to capture “oil price shocks” (such as net oil price increases) effectively identify the
exogenous component of the price of oil. This is not the case:
● Oil price shocks may occur in the absence of exogenous oil events.
● Exogenous oil events are not necessarily followed by oil price shocks.
● The exogenous oil events of 1973/74, 1978/79 and 2002/03 were followed by price shocks in nonoil industrial commodities. No such price shocks occurred in 1980/81 or 1990/91.
3. We can, however, assess the predictive power of exogenous oil supply shocks for changes in the
real price of oil.
How Well Do Exogenous Oil Supply Shocks Predict
Changes in the Real Price of Oil?
1. None of the existing measures of exogenous oil supply shocks explains the oil price data well.
2. Alternative explanations:
● The unexplained variation in oil prices in 1973/74, 1979/80 and 2003/04 can be explained by
shifts in the demand for oil.
● The unexplained variation in oil prices in 1990/91 can be attributed to shifts in uncertainty about
Saudi oil supplies.
● The 1980/81 oil price increases is well-explained by exogenous oil supply shocks, as defined in
this paper.
Instrumental Variable Regressions for U.S. Real GDP Growth
Regressand:
Δgdpt
Exogenous oil supply shocks measured as quantitative dummies
Exogenous oil supply shocks as
proposed in this paper
1973.I-2004.III
Δnptoil−1
1947.II2001.III
(1)
(2)
0.95
0.78
0.20
0.22
0.11
0.12
-0.04 -0.04
-0.17 -0.18
-0.03
-
1947.II2004.III
(3)
(4)
0.87
0.73
0.20
0.21
0.13
0.14
-0.05
-0.05
-0.15
-0.15
-0.02
-
1973.I2004.III
(5)
(6)
0.68
0.54
0.20
0.20
0.12
0.13
0.02
0.03
-0.07
-0.07
-0.02
-
1973.I2004.III
(7)
(8)
0.74
0.58
0.16
0.17
0.06
0.07
0.03
0.05
-0.06
-0.06
-0.02
-
(9)
0.57
0.17
0.11
0.11
-0.09
0.03
(10)
0.57
0.21
0.08
0.09
-0.08
-
(11)
0.56
0.23
0.03
0.06
-0.01
-
(12)
0.40
0.15
0.28
0.18
-0.14
-
(13)
0.49
0.19
0.08
0.06
-0.00
-
Δnptoil− 2
-0.05
-
-0.05
-
-0.04
-
-0.05
-
-0.09
-
-
-
-
Δnptoil−3
-0.01
-
-0.00
-
-0.00
-
-0.01
-
0.06
-
-
-
-
Δnptoil− 4
-0.06
-
-0.07
-
-0.03
-
-0.04
-
-0.01
-
-
-
-
Δrptoil−1
-
-0.03
-
-0.03
-
-0.02
-
-0.03
-
0.05
0.06
-0.04
-0.01
oil
t −2
Δrp
-
-0.06
-
-0.06
-
-0.04
-
-0.06
-
-0.08
-0.06
-0.12
-0.06
Δrptoil−3
-
-0.02
-
-0.01
-
-0.01
-
-0.02
-
0.07
0.04
0.06
-0.02
Δrptoil− 4
F test
(p-value)
-
-0.07
-
-0.07
-
-0.04
-
-0.04
-
-0.02
-0.01
0.03
-0.01
2.95
(0.02)
2.43
(0.05)
2.83
(0.03)
2.39
(0.05)
1.74
(0.15)
1.46
(0.22)
1.87
(0.12)
1.443
(0.225)
0.37
(0.83)
0.33
(0.86)
0.45
(0.77)
0.23
(0.92)
0.77
(0.55)
Regressors:
c
Δgdpt −1
Δgdpt − 2
Δgdpt −3
Δgdpt − 4
Notes: The instruments include a constant, four lags of real GDP growth and eight lags of the exogenous oil supply shock. Columns
(1)-(6) are based on the PPI for domestic crude oil as reported by the BLS and Hamilton (2003); columns (7)-(13) are based on the
average price of imported crude oil as reported by the U.S. Department of Energy, extended backwards from 1974.I to 1973.I as in
Barsky and Kilian (2004). The F-test refers to the null that oil prices changes have no effect on real GDP growth. Column (11)
excludes the Saudi production response; column (12) drops the embargo; column (13) includes Saudi Arabia in the benchmark starting
in 1974.
Dynamic Effects of a 10% Oil Price Increase: United States
IV Point Estimate with One-Standard Error Band and Two-Standard Error Band
4
Quarterly Real GDP Growth
3
2
1
0
-1
-2
-3
-4
1
2
3
4
5
Quarters
6
7
8
Weak Instrument Problems?
● Weak instruments produce biased IV estimators and hypothesis tests with large size distortions
(see Stock, Wright and Yogo (2002) for a review).
● Stock and Yogo (2003) propose a test of the null of weak instruments against strong instruments.
Result:
All IV regressions in the previous table fail this weak instrument test!
Dynamic Effects of a 10% World Oil Supply Disruption on United States
Direct Estimate Based on Quantitative Dummies
1
Real GDP Level
0
0
-1
-2
-5
-10
-3
-4
1
Quarterly CPI Inflation
5
2
3
4
5
6
7
8
9
10
11
-15
1
12
3
15
2
10
1
5
CPI Level
Quarterly Real GDP Growth
2
0
-5
-2
-10
2
3
4
5
6
7
Quarters
8
9
10
11
12
3
4
5
2
3
4
5
6
7
8
9
10
11
12
6
7
8
9
10
11
12
0
-1
-3
1
2
-15
1
Quarters